CN1837745A - Measurement system and method for reducing the effects of nonrandom noise - Google Patents

Abstract

The invention relates to a testing system for reducing non-random noise effect, and relative method, wherein said system comprises: a memory unit for storing at least one middle image, and a processor connected to the memory unit to fix the spatial relation between the interested area of middle image and the relative interested area of at least one reference image, and generate image to respond the spatial relation and several middle images; said middle image can comprises several parts and several interested areas.

Description

Measurement system and method for reducing the effects of nonrandom noise

related application

This application is a nonprovisional application of, and claims priority to, U.S. Provisional Application No. 60/615,906, filed October 4, 2004, which is hereby incorporated by reference.

field of invention

The invention relates to a measurement system and a method for improving the accuracy of the measurement system.

Background technique

Integrated circuits are very complex devices consisting of multiple layers. Various layers may contain conductive and insulating materials, while other layers may contain semiconducting materials. These various materials are arranged in patterns, usually according to the desired function of the integrated circuit. The pattern also reflects the manufacturing process of the integrated circuit.

Integrated circuits are manufactured through complex multi-level manufacturing processes. In this multi-stage process, resistive material is (i) deposited on a base layer, (ii) revealed by a photolithographic process, and (iii) revealed to create a pattern defining certain areas that are later etched.

Various measurement, inspection, and failure analysis techniques developed for inspection during assembly stages, integrated circuits between successive manufacturing stages, integrated with the manufacturing process (also referred to as "in-line" inspection techniques), or not combined (also referred to as "offline" detection techniques). Various optical and charged particle beam inspection tools and inspection tools are known in the art, such as VeraSEW ^™ , Compluss ^™ and SEMVision from Applied Materials, Inc. of Santa Clara, California (Applied Materials, Inc. of Santa Clara, California). ^TM .

Manufacturing failures can affect the electrical performance of integrated circuits. Some of these failures arise from unwanted deviations from the required pattern dimensions. A "critical dimension" is typically the width of a patterned line, the distance between two patterned lines, the width of a contact, or the like.

One of the measurement purposes is to determine whether the inspected structural component contains deviations from these critical dimensions. This inspection is usually performed by charged particle beam imaging, which provides the high resolution required to measure the deviations.

Measurement systems are highly complex. They include several mechanical components and detectors of the present technology. Various mechanical parts of the system, including but not limited to pumps, tables, etc. introduce noise. Noise is also introduced during the illumination and detection phases of the measurement process, for example due to fluctuating electromagnetic fields in the measurement system. This noise can reduce the resolution of the measurement system.

By summing multiple intermediate images, the effect of random noise is substantially reduced. Each intermediate image is produced by scanning an area with a charged particle beam and detecting electrons scattered from the area. The sum operation averages out the noise.

Non-random noise such as but not limited to repetitive noise affects the final image. These noises cause the intermediate images to shift relative to each other.

When the intermediate images are summed, this offset can distort the final image, since pixels belonging to different intermediate images at different locations in the area are summed to provide the final image.

The importance of resolution-enhancing techniques grows as the dimensions of submicron integrated circuit structures continue to shrink.

There is an increasing need to provide effective methods and systems for reducing the impact of noise.

Contents of the invention

A method comprising: determining a spatial relationship between regions of interest within at least one intermediate image and between corresponding regions of interest within at least one reference image; wherein the intermediate image comprises a plurality of intermediate image portions, and wherein The plurality of intermediate image portions includes a plurality of regions of interest; and generating an image responsive to the spatial relationship and responsive to the content of the plurality of intermediate images.

Advantageously, generating includes performing branch image alignment in response to spatial relationships.

According to an embodiment of the present invention, the method further includes generating a reference image. Advantageously, the generation of the reference image comprises scanning the reference area. Advantageously, the generation of the reference image reflects the content of at least one intermediate image.

Advantageously, the method further comprises determining at least one noise characteristic.

Advantageously, the method further comprises defining a region of interest, this stage may comprise searching for regions comprising edge information.

Advantageously, the stage of generating the image comprises adding pixels substantially representing the same position of a certain area.

A system comprising a storage unit adapted to store at least one intermediate image; and a storage unit connected to the storage unit and adapted to determine a corresponding interest between regions of interest within the at least one intermediate image and within at least one reference image A processor of a spatial relationship between regions; wherein the intermediate image includes a plurality of intermediate image portions, and wherein the plurality of intermediate image portions includes a plurality of regions of interest; and generating an image responsive to the spatial relationship and responsive to content of the plurality of intermediate images.

Advantageously, the processor is adapted to perform at least one of the following: (i) perform branch image alignment in response to spatial relationships; (ii) generate reference images; (iii) generate reference images in response to informational representations of reference regions; (iv) generate a reference image in response to the content of at least one intermediate image; (v) determine at least one noise signature; (vi) define a region of interest; (vii) search for regions containing edge information; Pixels at the same location in a certain area.

Advantageously, the system further comprises at least one detector for detecting the beam of charged particles scattered from the region.

Description of drawings

In order to understand the invention and see how it can be practiced in practice, a preferred embodiment will now be described, by way of non-limiting examples only, with reference to the accompanying drawings, in which:

1 is a schematic diagram of a critical dimension scanning electron microscope according to an embodiment of the present invention;

Fig. 2 shows a reference image according to an embodiment of the present invention;

Fig. 3 shows another reference image according to an embodiment of the present invention;

Figure 4 shows a horizontally derived image according to an embodiment of the present invention;

FIG. 5 shows a vertically derived image 360 according to an embodiment of the present invention;

Fig. 6 shows a method for generating an image according to an embodiment of the present invention;

Fig. 7 shows a method for generating an image according to an embodiment of the present invention;

Figure 8 shows an exemplary noise spectrum.

Detailed ways

A typical CD-SEM includes an electron gun for generating the electron beam, a deflection and tilt unit, and a focusing lens that can be used to scan the sample with the electron beam while reducing various aberrations under certain tilt conditions and dysregulation. Electrons, such as secondary electrons, which are neglected due to the interaction of the sample and the electron beam, are attracted to a detector which provides a detection signal which is processed by a processing unit. The detection signal can be used to determine various characteristics of the sample, as well as to form an image of the sample being inspected.

The present invention can be implemented on CD-SEMs of various configurations that differ from each other in the number of parts and the arrangement of the parts. For example the number of deflection units and the specific configuration of the individual units may vary. A CD-SEM can include in-lens as well as out-of-lens detectors or a combination of both.

For example, the measurement system of CD-SEM 100 is shown in Figure 1. The CD-SEM 100 includes (i) a scanner, such as a scanning deflection unit 102, for scanning an area comprising multiple structural features using a charged particle beam; (ii) at least one detector, such as an in-lens detector 16, which is fixed To receive charged particles from the interaction between the region and the charged particle beam and provide a plurality of detection signals; the detection signals from repeated scanning of the region can form an intermediate image; (iii) a storage unit, such as storage unit 9, is suitable for storing at least one intermediate image; and (iv) a processor, such as processor 8, connected to the storage unit, adapted to (a) determine between regions of interest within at least one intermediate image range and within at least one reference image range The spatial relationship between the corresponding regions of interest within; although the intermediate image includes multiple intermediate image parts, although the multiple intermediate image parts contain multiple regions of interest; and (b) in response to the spatial relationship and in response to multiple The content of the intermediate image to generate the image.

If the CD-SEM 100 contains multiple detectors, it can use the re-imaging registration results obtained by processing intermediate images from one processor to perform re-registration of images from other detectors.

According to an embodiment of the invention, only one detector's intermediate image is processed to estimate the spatial relationship between regions of interest, and the result is used to generate images of this detector as well as other detectors.

According to another embodiment of the invention, intermediate images of multiple detectors are processed and the spatial relationship between regions of interest associated with different detectors is processed to obtain a coincidence curve for one or more detectors.

CD-SEM 100 also comprises: (a) the electron gun 103 that emits electron beam 101, and this electron gun 103 is drawn out by pole 104, (b) the objective lens 112 that electron beam is focused on the surface 105a of inspected object 105, (c) deflection unit 108-111, and (d) a stage 101 for guiding relative mechanical motion between the object 105 and the objective lens 112.

Using the scanning deflection unit 102, the beam is scanned over the sample. Alignment of the aperture 106 to the beam or an ideal optical axis can be achieved using the deflection units 108 to 111 , respectively. As deflection unit coils, electrostatic modules of the charged disk type or combinations of coils and electrostatic deflectors can be used.

The in-lens detector 16 is capable of detecting secondary electrons with relatively low energy (3 to 50 eV) escaping from the object 105 at various angles. Measurements of scattered or secondary particles from the sample can be performed using a scintillator-style detector connected to a photomultiplier tube or the like. Since the method of measuring the signal generally does not affect the inventive concept, this is not to be understood as a limitation of the invention. It should be understood that a CD-SEM may additionally or alternatively comprise at least one off-lens detector.

The detection signals are processed by a processor 8, which may also be adapted to control the components of the CD-SEM 100 and regulate its operation. Advantageously, the processor 102 has image processing capabilities and can process the detection signal in various ways. A typical processing scheme involves generating a waveform representing the amplitude of the detected signal versus scan direction. The waveform is further processed to generate images, determine the location of at least one edge and other cross-sectional characteristics of the inspected structural component.

The different parts of the system are connected to corresponding power supply units (such as the high voltage power supply unit 21 ) controlled by various control units, most of which are omitted in the figure for simplicity of description. The control unit determines the current and voltage supplied to a component.

The CD-SEM 100 comprises a dual deflection yoke comprising deflection units 110 and 111. Thus, the beam tilt introduced in the first deflection unit 110 can be corrected for use in the second deflection unit 111 . Thanks to this double deflection system, the electron beam can be deflected in one direction without beam inclination of the incoming electron beam with respect to the optical axis.

FIG. 2 shows a reference image 200 according to an embodiment of the present invention. The reference image 200 has a rectangular shape. It is divided into a plurality of bars 201-209, each about 10 pixels high. A plurality of regions of interest 211-216 are defined in the image. The region of interest is defined, for example, to contain edge information of the measuring means, for example contact points. The remaining edge information can be ignored.

Note that several bars may contain spatially continuous lines or spatially discontinuous lines. The strip usually consists of lines scanned in a continuous fashion. For example, in an interlaced scanning manner, where the first scan part includes the first, (q+1) ^th , (2q+1) ^th , ... (nq+1) ^th lines, and the subsequent scan part includes the second, (q+ 2) ^th , (2q+2) ^th , ...(nq+2) ^th lines, etc., the first one contains the first to q ^th lines, the second one contains the second to 2k ^th lines, and so on.

Advantageously, each region of interest contains edge information. Edge information can be detected using various prior art schemes, such as tracking after first, second and/or third derivatives of the image. The derivative can be an x-axis derivative, a y-axis derivative, a binary derivative, but this is not required.

Advantageously, the region of interest is selected by finding branch regions that contain the largest amount of edge information, above a threshold and the like. The effective value of the edge information can be calculated in many ways, such as the gray level difference between adjacent pixels, the number of different pixels, and so on.

FIG. 3 shows another reference image 300 , FIG. 4 shows a horizontal derivative image 330 , and FIG. 5 shows a vertical derivative image 360 , according to various embodiments of the invention.

Reference image 300 includes an image of vertical line 302 and an image of additional structure 304 . The vertical line image 302 and the additional feature image 304 include substantially edge information. A plurality of regions of interest 310 are defined, for example to contain edge information. The position of the region of interest can be determined by locating edge information. This can be done by generating images of the horizontal and/or vertical derivatives.

It should be noted that the present invention can be used for images having various shapes and different bands of image parts. Also note that the number of pixels per stripe is different from ten.

Advantageously, the size, especially the width, of the stripes responds to various parameters including noise frequency, beam scanning rate and Nyquist's law. The present invention is very effective if the characteristic frequency of the noise is substantially lower than the frequency of the line scan of the SEM image.

Assume that at T1 scanning line, the number of lines in the image is Ni, the number of lines in the strip is Ns, and the characteristic frequency of noise (such as the highest frequency containing substantial noise energy) is Fn, then: 2*Fn ≤1/(T1*Ns), then, 1/(2*Fn*T1)≥Ns.

The present invention uses a CD-SEM to scan an image comprising 480 lines at a frequency of 75 Hz, the characteristic frequency of noise is 930 Hz, and the present invention uses a strip of ten lines.

According to various embodiments of the present invention, pixels of certain intermediate images of certain regions are compared with reference pixels of a reference image. The reference image may be a previous intermediate image of substantially the same area, a combination of one or more previous intermediate images of the same area, an intermediate image of the reference area, a combination of one or more intermediate images of the reference area, a reference The final image of the region, the composite image of the region, the composite image of the reference region, or a combination of any one or more of these images.

According to an embodiment of the present invention, the reference image may be obtained using different imaging/scanning conditions for obtaining the intermediate images. To provide higher quality reference images, higher currents, lower scan rates, etc. can be used.

The reference area usually contains objects (also referred to as measurement models), such as visible pointing devices, lines, etc. Targets are typically selected in response to characteristics measured during the measurement process. For example, if the width of a line is being measured, then the target is usually ideally equal to the line being measured. If for example vias are measured, the target should be vias.

Advantageously, the target image (reference image) contains substantially edge information. This edge information is conveniently located within slices of the reference image.

Advantageously, a relative motion is calculated between each strip and the corresponding reference strip and used to achieve a strip-based (also called branch image) registration.

Advantageously, the invention also includes performing noise spectrum analysis. During unique time periods, individual stripes are scanned. Relative motion can be thought of as sampling along the time axis. These temporal samples may provide a noise spectrum by performing a time-to-frequency transformation such as, but not limited to, a Fourier transform. For example, assume that at time T0, scanning of the first strip of the first intermediate image ends. At time T0+Ns*T1, the scanning of the second strip of the first intermediate image ends. Generally, the scanning of the k-th slice of the j-th intermediate image ends at time T0+(Ns*k+Ni*j)*T1. These samples can be transformed into the frequency domain by applying a discrete Fourier transform.

The relative motion between the slices of the currently acquired intermediate image and the corresponding slices of the reference image can be set in the form of a vector which undergoes a time-to-frequency domain transformation.

According to one embodiment of the invention, relative horizontal motion between slices belonging to one or more intermediate images and a reference image is calculated and transformed to provide an x-axis noise spectrum.

According to one embodiment of the invention, relative horizontal motion between slices belonging to one or more intermediate images and a reference image is calculated and transformed to provide a y-axis noise spectrum.

According to another embodiment of the invention, relative horizontal motion between slices belonging to one or more intermediate images and a reference image is calculated and transformed to provide a binary noise spectrum.

Advantageously, by applying branch frame coincidence, the pixels summed along the intermediate image (frame) will correspond to the same point of the scanned area (sample).

FIG. 6 shows a method 400 for generating an image according to an embodiment of the present invention.

Method 400 begins with stage 410 of generating or receiving a reference image. According to one embodiment of the invention, a reference image may be obtained before the measurement process starts. The reference image can be synthesized from an electronic description of the integrated circuit or one of the layers of the integrated circuit.

A reference image may be generated by scanning a reference area containing a reference target, performing branch frame alignment and providing a reference image.

Stage 410 is followed by stage 420 of obtaining a plurality of intermediate images of an area. This area includes the features being measured during the measurement. This characteristic may be, for example, the width, length or diameter of a structure, the distance between two structures, and similar quantities. A data description (eg pixel values) of the intermediate image can be stored for subsequent stages of the method.

Stage 420 is followed by stage 430 of determining spatial relationships between regions of interest in at least one intermediate image and corresponding regions of interest in at least one reference image.

Advantageously, the intermediate image contains a plurality of intermediate image parts, and the plurality of intermediate image parts contains a plurality of regions of interest.

Advantageously, relative movement between adjacent intermediate images is relatively limited. The motion can be calculated by comparing the positions of the edge information within the corresponding regions of interest. The search is limited to a relatively small area (about 2 pixels of motion in each direction, but this is not required). The search may include performing a correction operation between a region of interest in an intermediate image and a corresponding region of interest in a reference image.

Various filters including NGC based filters can be applied during this search.

According to one embodiment of the invention, the search includes resampling of the slice and the corresponding reference slice, the search for edge information, and once it is found, is performed by using the original image or a resampled image with a resampling factor More precise determination of spatial relationships.

For example, a reference image may include 100x100 pixels, while an intermediate image may include 480x480. A 50x50 reference image can be compared to a 240x240 intermediate image by performing a resampling by a factor of 2 (in each dimension). This reduces the complexity of the search.

According to one embodiment of the invention, the intermediate image may be smoothed before initiating the search for the spatial relationship between the intermediate image and the reference image.

Once the relative motion is defined, the method can generate registration curves, including smooth registration curves using registration offset arrays, block confidence arrays, expected noise frequencies, and the like. This confidence level can help to ignore data obtained with a low confidence level. The confidence level can be derived from the amount of edge information contained within the slice, etc.

Tracking techniques such as applying a Kalman filter can also be used to define coincidence curves.

Coincident curves can be used to offset individual lines or groups of lines in the intermediate image. Thus, although the relative movements of the individual strips are calculated, these movements can be processed to produce a coincident curve according to which the individual lines will be offset within the individual strips, in other words, lines belonging to the same strip can be offset from each other. Different ways are offset.

The final image is produced by summing the pixels of the portion of the image affected by the coincident curve.

Stage 430 is followed by stage 440 of generating images in response to spatial relationships and in response to the content of the plurality of intermediate images.

Advantageously, stage 440 includes performing branch image registration in response to spatial relationships.

This image can be generated by adding pixels (belonging to a different intermediate image) corresponding to approximately the same position of the region after branch image registration.

Depending on the detected spatial relationship, stage 440 may include offsetting (or wrapping) the strips. This wrapping can cause overlaps and gaps. Pixel values within the gap can be calculated by interpolation. The gap in the overlapping portion is usually the sum of pixels of multiple offset strips.

According to various embodiments of the invention, a plurality of spatial relationships are provided to a filtering algorithm to determine the coincidence curves of the strips used to form the respective intermediate images.

Stage 440 is preferably followed by stage 450 of determining at least one noise characteristic.

Stage 440 may also be followed by stage 460 of processing the image to determine at least one critical dimension.

Advantageously, the reference image is a measurement model. The measurement model includes the target image as well as unimportant additional information. This measurement model is shown in FIG. 2 .

The measurement phantom 240 is imaged to estimate various parameters of the contact point placed in the center of the phantom image. It should be noted that the reference image can be generated by using various algorithms such as Normalized Gray Level Correction (NGC) technique. This algorithm is quite robust to noisy data and can provide continuous branch image coincidence for a single scan frame.

FIG. 7 illustrates a method 800 for generating an image in accordance with one embodiment of the present invention.

Method 800 differs from method 400 in the manner in which reference images are generated. In this method, an intermediate image is generated by processing the region containing the measured features. Thus, instead of using a reference image of another reference area, during the acquisition of the intermediate image, a reference image is constructed. It should be noted that the generation of reference images can be realized after all intermediate images have been acquired.

The first or further intermediate images thus form the reference image. Other intermediate images are compared with the reference image to obtain relative motion between adjacent intermediate images.

The method 800 is initiated by a stage 420 of obtaining at least one intermediate image of an area.

Stage 420 is followed by stage 425 of generating a reference image from at least one intermediate image. Stage 425 may be followed by stage 428 of obtaining an auxiliary intermediate image. Advantageously, stage 428 may include updating the reference image.

Stage 428 is followed by stage 430 of determining spatial relationships between regions of interest within at least one intermediate image and corresponding regions of interest within at least one reference image.

Stage 430 is followed by stage 440 of generating images in response to spatial relationships and in response to the content of the plurality of intermediate images.

Advantageously, stage 440 includes performing branch image registration in response to spatial relationships.

Stage 440 is preferably followed by stage 450 of determining at least one noise characteristic. Figure 8 shows an exemplary noise spectrum measured using the present invention.

Stage 440 may also be followed by stage 460 of processing the image to determine at least one critical dimension.

According to an embodiment of the present invention, each intermediate image serves as a reference image for the next desired intermediate image.

According to another embodiment of the invention, the reference image is obtained by applying the various stages of the method 800, and the image generated during stage 440 is the reference image. After the reference image is acquired, images of the area or other areas may be acquired using methods 800 or 400 (while utilizing the reference image).

The invention can be implemented using conventional tools, measurements and components. Accordingly, the details of such tools, components and measurements are not elaborated here. In the foregoing description, numerous specific details are set forth, such as CD-SEM structures, images and stripe shapes, etc., in order to provide a thorough understanding of the present invention. It should be realized, however, that the invention can be practiced without the foregoing specific details.

Only exemplary embodiments of the invention and only a few general examples thereof have been shown and described in this specification. It is to be understood that the present invention is capable of employing various other combinations and configurations and is capable of changes or permutations within the scope of the inventive principles as expressed herein.

Claims (19)

1. A method comprising: determining a spatial relationship between regions of interest within at least one intermediate image and between corresponding regions of interest within at least one reference image; wherein the intermediate image includes a plurality of intermediate image portions, and wherein the plurality of intermediate image portions includes multiple regions of interest; and Images are generated responsive to spatial relationships and to the content of multiple intermediate images. 2. A method according to claim 1, wherein the generating includes performing branch image registration in response to spatial relationships. 3. The method according to claim 1, further comprising generating a reference image. 4. A method according to claim 3, wherein the generation of the reference image comprises scanning the reference area. 5. A method according to claim 3, wherein the generation of the reference image reflects the content of at least one intermediate image. 6. The method according to claim 1, further comprising determining at least one noise characteristic. 7. The method according to claim 1, further comprising defining a region of interest. 8. A method according to claim 7, wherein the determining stage comprises searching for regions comprising edge information. 9. A method according to claim 1, wherein the stage of generating the image comprises adding pixels representing substantially the same location of an area. 10. A system comprising a storage unit adapted to store at least one intermediate image; and a storage unit connected to the storage unit adapted to determine corresponding sensory differences between regions of interest within at least one intermediate image and within at least one reference image A processor of a spatial relationship between regions of interest; wherein the intermediate image comprises a plurality of intermediate image portions, and wherein the plurality of intermediate image portions comprises a plurality of regions of interest; and generating an image responsive to the spatial relationship and responsive to the content of the plurality of intermediate images . 11. The system according to claim 10, wherein the processor is adapted to perform branch image registration in response to the spatial relationship. 12. A system according to claim 10, wherein the processor is adapted to generate the reference image. 13. A system according to claim 12, wherein the processor is adapted to generate the reference image in response to the information representation of the reference area. 14. A system according to claim 12, wherein the processor is adapted to generate the reference image in response to the content of the at least one intermediate image. 15. The system according to claim 10, wherein the processor is adapted to determine at least one noise characteristic. 16. The system according to claim 10, wherein the processor is adapted to define a region of interest. 17. A system according to claim 16, wherein the processor is adapted to search for regions containing edge information. 18. A system according to claim 10, wherein the processor is adapted to add pixels representing substantially the same location of an area. 19. The system according to claim 10, further comprising at least one detector for detecting the beam of charged particles scattered from the region.

Images